In urodeles and anurans a major component of retinal regeneration is transdifferentiation of RPE to neural retina. But regeneration is limited in mammals, leaving them susceptible to retinal injury and blinding diseases. Recent studies identified a population of RPE stem cells (RPESC) among cultures of human RPE. These RPESC show unrestricted proliferation and their differentiation potential closely resembled mesenchymal stem cells (MSC), which share a common neuroepithelial origin with RPE. Although RPESC differentiated to express neuronal markers, they failed to induce photoreceptor markers such as RHO. Thus, these RPESC may not represent an intermediate in transdifferentiation of mammalian RPE into photoreceptors. We examined cultures of adult mouse RPE for cells with the differentiation capacity of RPESC, but we failed to identify such cells. However, we found that cells with properties of RPESC can be efficiently and stably induced from RPE (iRPESC) through a hypoxia-dependent pathway, similar to that described for maintenance and induction of MSC. Hypoxia causes RPE damage and is linked to neovascularization and AMD. Key to this iRPESC reprogramming pathway is superinduction of hypoxia inducible factor 1a (Hif1a) to a threshold sufficient to bind and activate the Oct4 stem cell gene promoter. Oct4 in turn induces Dnmt1 which silences cell cycle blocking cyclin dependent kinase inhibitors leading to unrestricted proliferation. These iRPESC are resistant to hypoxia, and importantly, as opposed to human RPESC, they differentiate into Rho+ cells-indeed this differentiation to Rho+ cells is more efficient than seen with embryonic stem cells or induced pluripotent stem cells. Furthermore, iRPESC do not undergo the typical epithelial-mesenchymal transition (EMT) seen when RPE are placed in culture, providing the potential for retaining an RPE phenotype as the cells are expanded. Because blinding diseases such as AMD are highlighted by loss of both functional RPE and photoreceptors, the ability of the iRPESC to undergo photoreceptor differentiation and to resist EMT-initiated loss of phenotype suggest a unique therapeutic potential for the cells. During the two year period of this R21 proposal, we aim to investigate the iRPESC reprogramming pathway on a molecular level. The purpose of these studies is to provide a foundation for experiments designed to optimize photoreceptor differentiation from iRPESC and to maintain a function RPE phenotype as iRPESC are expanded, so in the future we can begin testing the effectiveness of the differentiated iRPESC in transplantation experiments. A second point of this molecular analysis is to identify pathway markers that can ultimately be used for detection of iRPESC in vivo, and to understand factors that might be used in the future to stimulate iRPESC generation from RPE in situ.
Retinal pigment epithelium (RPE) can efficiently transdifferentiate to replace damaged neural retina in newts and frogs, and to a more limited degree in embryonic chick and rodents. However, RPE transdifferentiation does not occur in adult mammals, and regeneration is limited leaving mammals susceptible to retinal injury and blinding diseases. Recent studies have provided the first evidence of RPE stem cells (RPESC) in cultures of human RPE. This important finding raises the possibility that these RPESC may serve as a reservoir for replacement of damaged RPE in diseases such as AMD, and they may represent an intermediate in transdifferentiation of RPE to photoreceptor in mammals. However, even though these human RPESC closely resembled mesemchymal stem cells (MSC) in their differentiation potential and they showed evidence of neuronal differentiation, they failed to induce markers of photoreceptor differentiation. We have examined adult mouse RPE for cells with properties of RPESC. Although we did not detect RPESC, we found instead that cells with properties of RPESC could be stably and efficiently induced from RPE (iRPESC) through a hypoxia-dependent pathway, similar to that described for maintenance and induction of MSC. Key to this iRPESC reprogramming pathway is superinduction of hypoxia inducible factor 1a (Hif1a) to a threshold sufficient to bind and activate the Oct4 stem cell gene promoter. Oct4 in turn induces Dnmt1 which silences cell cycle blocking cyclin dependent kinase inhibitors leading to unrestricted proliferation. These iRPESC are resistant to hypoxia, and importantly, as opposed to human RPESC, they differentiate into Rho+ cells-indeed this differentiation to Rho+ cells is more efficient than seen with embryonic stem cells or induced pluripotent stem cells. These results with iRPESC demonstrate induction of RPESC under hypoxic conditions that have been linked to RPE damage in retinal disease (e.g., neovascularization and AMD). Because iRPESC can efficiently differentiate to express rod photoreceptor makers, they can act as a stable intermediate for transition of mammalian RPE to cells expressing markers of photoreceptors. We propose studies of the molecular pathway leading to iRPESC reprogramming during the two year period of the R21.
